Choosing Pair-Trade Hedge Ratios: Regression Beta vs Dollar Neutrality
Summary
The document asks why pairs traders commonly size the two legs using a regression coefficient rather than equal dollar amounts. Regression estimates how one stock’s movement relates to the other and can define a mean-reverting spread; dollar neutrality instead balances the market value of the long and short positions. These approaches target different properties: spread construction from historical co-movement versus reduced exposure to broad market moves.
The question proposes using the regression-based spread to choose trade direction, then sizing both legs dollar-neutrally to limit market exposure. It does not provide an answer, empirical comparison, or evidence that this alternative performs better. The practical trade-off is left unresolved: dollar neutrality may not preserve the regression relationship that motivates the spread, while a regression hedge can leave net market exposure. The document is useful as a framing of the hedge-ratio choice, but does not establish when either method is preferable.
Key ideas
- Regression hedge ratios use estimated relative stock movements to define a pair spread.
- Dollar-neutral sizing balances the market value of the long and short legs.
- A regression hedge ratio need not produce dollar neutrality.
- The document raises but does not resolve whether dollar-neutral sizing can retain spread signals while reducing broad market exposure.
Tags
Full text
# Pair trading: Hege ratio by price ratio or by regression on the stocks # Pair trading: Hege ratio by price ratio or by regression on the stocks So I feel things like this question do exist on the site however I failed to form a conclusion based on what was written, so I decided to formulate my query as exactly as possible and hopefully anyone else as confused as I am will find an answer here and be able to go from this page enlightened - with this is mind: I understand that pair trading relies on the cointegration of 2(or more) stocks. They tend to maintain a certain position relative to each other probabilistically maintaining a certain 'spread'. By regressing one stock relative to the other you can attain a beta value often used in pair trading algorithms, which is used to define said spread. This regressed beta seems to be often used to indicate how much one should buy of one stock relative to the other. I am also aware that if one were to hedge in a dollar neutral fashion by going long on one stock as much as one has gone short on the other (in dollar terms) then one is resilient to overall market movement. My question is that the former seems to be far more common than the latter in pair trading and I wanted to know why? At a glance it would seem that the regressed beta gives you the amount one stock moves relative to the other in general but - provided that this beta didn't happen to equal the ratio of one price to the other - one would still remain exposed to large market movements being either more long or short in dollar terms. The dollar neutral hedge ratio would not represent the ratio of how much the stocks move relative to each other but one can still use the spread(with this regressed beta which here doesn't get used to choose how much we buy of one stock or the other), establish one should go long or short on each stock in the pair respectively, and then one could bet in a dollar neutral fashion giving market movement protection. The second question is why is this not done more?
Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.